The Semantics of yue yue in Mandarin Chinese 1 1 2 Xiao Li and Carlos A. Fasola 2 Queens College, CUNY, Rutgers University In this paper, we argue that yue yue in Mandarin Chinese can mark two semantically distinct comparative structures: comparative correlatives and adverbial comparatives. Comparative correlatives are sentences where the first copy of yue precedes a gradable predicate (typically adjectives), and adverbial comparatives are sentences where the first copy of yue precedes a non-gradable predicate (typically verbs). Comparative correlatives and adverbial comparatives are truth-conditionally distinct. The latter has an obligatory temporal reading absent in the former. Based on the semantic difference of these two types of yue yue comparatives, we argue that gradable predicates (typically adjectives) contain a degree argument in their semantics, but lack a time argument; nongradable predicates (typically verbs) have a time argument, but lack a degree argument. 1.Introduction It has been observed that sentences in Mandarin Chinese marked by the form of yue yue, with yue preceding either an adjective, e.g. (1a), or a verb, e.g. (1b), correspond to so-called comparative correlatives in other languages, such as the English translations (Chao 1968, Li and Thomas 1981, Hsiao and Tsao 2002, Lin 2007, Liu 2008). (1) a. Pingguo yue da yue tian. Apple big sweet The bigger an apple is, the sweeter it is. b. John yue xihuan Mary, Jane yue gaoxing. like happy The more John likes Mary, the happier Jane is. However, it has rarely been noticed that when the first yue precedes a certain class of predicates, which we characterize as non-gradable predicates, such as pao run in (2a), the sentence receives a different interpretation from typical comparative correlatives and instead receives an interpretation like so-called adverbial comparatives, as in the English translations in (2).
(2) a. John yue pao yue kuai. John run fast John ran faster and faster. b. John yue chang ge, xinqing yue hao. John sing songs mood good As John was singing, his mood became better and better. The goal of this paper is two-fold. First, we want to establish that comparative correlatives like (1) and adverbial comparatives like (2) are truth-conditionally distinct. The latter has an obligatory temporal interpretation absent in the former. This will be discussed in detail in section 2. Second, we will propose an analysis which captures the difference in semantic content between comparative correlatives and adverbial comparatives. Our analysis crucially refers to the distinction between gradable and non-gradable predicates. Comparative correlatives have the first occurrence of yue appearing in front of a gradable adjective, e.g. da big in (1a), or a gradable verb, e.g. xihuan like in (1b). Adverbial comparatives, on the other hand, have the first yue occurring in front of a non-gradable verb, e.g. pao run in (2a) and chang sing in (2b).The gradability of a predicate can be decided by (i) whether it can be modified by a degree modifier such as hen very, e.g. (3) and (5), and (ii) whether it can be used directly in the bi-comparative1, e.g. (4) and (6). (3) a. John hen gao. very tall John is very tall. b. John hen xihuan zhongguo. very like China John likes China very much. (4) a. John bi Mary gao. tall John is taller than Mary. 1 The syntax and semantics of the bi-comparative have been studied in detail in Li and Thompson (1981), Liu (1996), Xiang (2003, 2005), Erlewine (2007), Lin (2009), Li (2009) and references therein. 37
b. John bi Mary xihuan Zhongguo. like China John likes China more than Mary does. (5) a. *John hen pao very run b *John hen chang ge. very sing song (6) a. *John bi Mary pao. run b. *John bi Mary chang ge sing song Based on the semantic difference between these two types of yue yue comparatives, we argue that gradable predicates (typically Adjectives) do not contain a time argument in their semantics while non-gradable predicates (typically Verbs) do, and, on the other hand, gradable predicates do contain a degree argument, while non-gradable predicates do not. 2. The semantic difference between Comparative correlatives and adverbial comparatives In this section, we show that comparative correlatives and adverbial comparatives are truth conditionally distinct. To begin with, let us consider the truth condition of a comparative correlative. A comparative correlative is true iff an increase of the degree of the property indicated by the predicate after the first yue is accompanied by an increase of the degree of the property indicated by the predicate after the second yue (Lin 2007, Liu 2008). For instance, the comparative correlative in (1a) is true iff an increase of an apple s size correlates with an increase of its sweetness. This meaning is illustrated by the scenario in (8a), in which (1a) is intuitively true. (8) a. The scenario in which (1a) is true Apples size Apple A: 6 cm in radius Apple B: 5 cm in radius Apple C: 4 cm in radius Apple s degree of sweetness A s sweetness: 10 B s sweetness: 7 C s sweetness: 5 38
(1a) is false if an increase of an apples size does not correlate with an increase of its sweetness, as shown by the scenario in (8b): (8) b. The scenario in which (1b) is false Apples size Apple A: 6 cm in radius Apple B: 5 cm in radius Apple C: 4 cm in radius Apple s degree of sweetness A s sweetness: 7 B s sweetness: 10 C s sweetness: 5 By the same token, (1b) means an increase of John s liking of Mary is accompanied by an increase of Jane s happiness. On the other hand, the truth condition of an adverbial comparative is different from the truth condition of a comparative correlative. An adverbial comparative is true iff the degree of the property indicated by the predicate after the second yue increases over time. For instance, the adverbial comparative in (2a) is true iff John s running speed increases over time. This meaning is illustrated by the scenario in (9a), where (2a) is intuitively true. (9) a. The scenario in which (2a) is true Temporally ordered running events 3 rd week of running 2 nd week of running 1 st week of running Average Speed His average speed was 6 mph His average speed was 5 mph His average speed was 4 mph (2b) is false if John s speed does not increase over time, as illustrated by the scenario in (9b): (9) b. The scenario in which (2a) is false Temporally ordered running events 3 rd week of running 2 nd week of running 1 st week of running Average Speed His average speed was 3 mph His average speed was 6 mph His average speed was 4 mph It s worth noting that in evaluating the truth value of (8b) in (9), we do not need to take 39
into consideration how many times that John ran, unlike what we did in comparative correlatives. All we need to know is whether his speed increases over time. The truth-conditional difference between comparative correlatives and adverbial comparatives can be further seen by comparing the near minimal pair of the comparative correlative in (10a) and the adverbial comparative in (10b). (10) a. John pao-de yue duo, ta (jiu) pao-de yue kuai. run-de much he (then) run-de fast The more John ran, the faster he went. b. John yue pao yue kuai. run fast John ran faster and faster. In (10a), the first copy of yue precedes a gradable adjective duo much. The sentence is based on the two non-comparative sentences John pao-de hen duo John ran a lot and John pao-de hen kuai John ran fast. Semantically, (10a) describes a correlation between the quantity of John s running and the speed he achieved. The example in (10b), repeated from (2a), is an adverbial comparative as the first copy of yue precedes the non-gradable verb pao run. Semantically, (10b) means that John s running speed increases over time. Let us compare the truth-values of (10a) and (10b) in the scenario described in (11): (11) Scenario: John did marathon training for 3 weeks. In the 1st week, John ran 7 times, and his average running speed was 6 mph. In the 2nd week, John ran 5 times, and his average running speed was 5 mph. In the 3rd week, John ran 3 times, and his average running speed was 4 mph. Time Number of Times Average Speed Week 3 Week 2 Week 1 John ran 3 times. John ran 5 times. John ran 7 times His average speed was 4 mph His average speed was 5 mph His average speed was 6 mph The comparative correlative in (10a) is ambiguous between two readings. On one reading, it says that the number of times that John ran each week (the second column in 11), correlates with his average speed per week (the third column in 11). Under this reading, (10a) is intuitively true in (11), because as the number of times that John ran per week decreases, his average running speed per week also decreases. Besides this reading, (10a) has another reading, according to which, (10a) means that there is a correlation between a running total of the number of times that John ran (the 40
second column in 11 ), and his average speed (the third column in 11 ). We will refer to this reading as the cumulative reading, and the previous reading as the non-cumulative reading. (11 ) Time Running Total Average Speed Week 3 + week 2 +week 3 Week 2 + week1 Week 1 John ran 3 + 5 + 7 times John ran 5 + 7 times. John ran 7 times His average speed was 5.2 mph His average speed was 5.5 mph His average speed was 6 mph Under the cumulative reading, (10a) is intuitively false, because as the total number of times that John ran increasing, his average speed decreases, as shown in the table in (11 ). Comparing (10b) to (10a), (10b) has only one reading, which expresses a correlation between John s running speed (the third column in 11), and time (the first column in 11). Intuitively (10b) is false in (9), because as time moves forward, John s running speed decreases. Let us refer to this reading as the temporal reading. From the examples in (10a) and (10b), we conclude that the temporal reading is not the same as the non-cumulative reading of comparative correlatives, as they do not yield the same truth value in the given scenario in (11). However, a question arises as to whether the temporal reading of adverbial comparatives is equivalent to the cumulative reading of comparative correlatives. If the answer to the question is yes, then this will invalidate the distinction that we have been trying to make between comparative correlatives and adverbial comparatives. In what follows, we will present two sets of evidence to show that the temporal reading of adverbial comparative is distinct from the cumulative reading of comparative correlatives. First, the cumulative reading of a comparative correlative is truth-conditionally weaker than the temporal reading of an adverbial comparative correlative. That is, the former can be true in scenarios where the latter is false. This is shown by the tables in (12) and (12 ). (12) Scenario: John did marathon training for 3 weeks. In the 1st week, John ran 3 times, and his average running speed was 6 mph. In the 2nd week, John ran 5 times, and his average running speed was 7 mph. In the 3rd week, John ran 3 times, and his average running speed was 6.7 mph. Time Number of Times Average Speed Week 3 Week 2 Week 1 John ran 7 times John ran 5 times John ran 3 times His average speed was 6.7 mph His average speed was 7 mph His average speed was 6 mph 41
According to the scenario in (12), the adverbial comparative in (10b) is intuitively false, because John s speed does not increase over time. However, the comparative correlative in (10a) is true in (12 ) under the cumulative reading, because with an increase of the total number of times that John ran, his average speed increases. (12 ) Time Running Total Average Speed Week 3 + week 2 +week 3 Week 2 + week1 Week 1 John ran 3 + 5 + 7 times John ran 5 + 3 times. John ran 3 times His average speed was 6.66 mph His average speed was 6.6 mph His average speed was 6 mph Second, not every adverbial comparative can be paraphrased by a comparative correlative. Let us look at the examples in (13) below: (13) a. (fan), John yue chi yue shao. rice eat few John ate less and less (rice). b. (huazhuang pin), Mary yue mai yue pianyi. cosmetics buy cheap Mary bought cheaper and cheaper cosmetics. c. yu yue xia yue xiao. rain fall little It was raining lighter and lighter. (13) are examples of adverbial comparatives. The first yue precedes a non-gradable verb, and the second yue precedes a negative adjective shao few (13a), pianyi cheap (13b), and xiao small (13c). The sentence in (13a) means: the degree of fewness of the quantity of the rice that John consumed increases over time, or, the quantity of rice that John consumed decreases over time. (13a) is intuitively true in a situation like the following: (14) Scenario: John is on a diet. On the 1st day, he ate 3 bowls of rice; on the 2nd day, he ate 2 bowls of rice; on the 3rd day, he only ate 1 bowl of rice. Time Day 3 Day 2 Day 1 Quantity of rice 1 bowl of rice 2 bowls of rice 3 bowls of rice 42
When we evaluate the truth value of (13a) in (14), we compare the quantity of the rice that John consumed in each temporally ordered eating event. If the quantity of rice that John ate decreases over time, then (13a) is true in (14), which is indeed the case here. (13a) cannot be paraphrased by a cumulative reading of a comparative correlative, because with the quantity of eating increasing, the rice consumed necessarily increases, as shown by the table in (14 ). (14 ) Time Quantity of rice Day 3 +Day 2 + Day1 Day 2 + Day1 Day 1 1+2+3 bowls of rice 2+3 bowls of rice 3 bowls of rice (13b) and (13c) illustrate the same idea. (13b) says that the price of the cosmetics that Mary bought in each buying situation decreases over time. Pianyi cheap is a property that applies to the cosmetics that Mary bought each time, instead of the total price she paid for all her buying. (13c) means that the volume of rain falling decreases over time. They both have a reading which cannot be paraphrased by the cumulative reading of a comparative correlative. Let us recap. In this section we have discussed the semantic difference between the two types of truth-conditionally distinct comparative structures marked by yue yue: adverbial comparatives and comparative correlatives. We have shown that the former has a temporal reading, which is lacking in the latter. In the following section, we will review Lin (2007) s analysis of yue yue comparatives. We show that as Lin s analysis of yue yue does not differentiate between comparative correlatives and adverbial comparatives, it fails to capture the semantic difference between them. 3. Lin (2007) s analysis of yue yue comparatives Lin (2007), following Beck (1997) s semantic analysis of English comparative correlatives, argues that yue yue constructions in Mandarin Chinese uniformly express a correlation between two pairs of degrees provided by the two subordinate clauses marked by yue. Let us take (15) as an example, and look at the details of his analysis. (15) Zhangsan yue shengqi, Lisi yue gaoxing. angry happy The angrier Zhangsan is, the happier Lisi is. Lin assumes that yue yue constructions have a quantificational structure like a conditional. He proposes that the sentence in (15) has the logical form in (16) 43
(16) CP CP 1 CP 2 yue IP 1 yue IP 2 NP AP NP AP Zhangsan shengqi Lisi gaoxing angry happy According to (16), (15) consists of two subordinate clauses CP 1 and CP 2. In each clause, yue takes a sentential complement IP. A covert universal quantifier takes both CP 1 and CP 2 as its semantic arguments. CP 1 provides the domain of quantification for and CP 2 provides a nuclear scope. The interpretations of some main components in (16) are provided in (17): (17) a. shengqi = x e d d s s angry (x)(d)(s) b. yue = P <d<s, t>> g 1 g 2 s 1 s 2 [P(g 1 )( s 1 ) P(g 2 )( s 2 ) g 2 g 1 ] c. = G< d, <d, <s, <s, t>>>> Q <d, <d, <s, <s, t>>>> g 1 g 2 s 1 s 2 [G(g 1 )(g 2 )(s 1 )(s 2 )] g 3 g 4 s 3 s 4 [Q(g 1 )(g 2 )(s 1 )(s 2 )] d. Zhangsan yue shengqi, Lisi yue gaoxing = g 1 g 2 s 1 s 2 [angry (Zhangsan)(g 1 )(s 1 ) angry (Zhangsan)(g 2 )(s 2 ) g 2 g 1 ] g 3 g 4 s 3 s 4 [s 1 s 3 s 2 s 4 R<<g 1, s 1 >,<g 3, s 3 >> R<<g 2, s 2 >, <g 4, s 4 >> happy (Lisi)(g 3 )(s 3 ) happy (Lisi)(g 4 )(s 4 ) g 4 g 3 ] (16d) reads as: For any pair of degrees g 1 and g 2, and any pair of situations s 1 and s 2 such that Zhangsan is angry to degree g 1 in s 1, and Zhangsan is angry to degree g 2 in s 2, and g2 is greater than g 1, there exists a pair of degrees g 3 and g 4, and a pair of situations s 3 and s 4 such that s 3 is an extended situation of s 1 and s 4 is an extended situation of s 2. Lisi is happy to degree g 3 in s 3, and Lisi is angry to degree g 4 in s 4. g 4 is greater than g 3. Moreover, g 1 in s 1 has a causative relation R relation with g 3 in s 3. g 2 in s 2 has a causative relation R relation with g 4 in s 4. In short, (17d) conveys the meaning that with an increase of Zhangshan s anger, there is an increase of Lisi s happiness. Although Lin s analysis successfully accounts for comparative correlatives like (15), his analysis does not extend easily to adverbial comparatives like (18). For one thing, it is a rather debatable claim that non-gradable verbs like pao run have a degree argument, just like gradable adjectives. In particular, in Mandarin Chinese, gradable and non- 44
gradable predicates can be clearly defined by whether they can be modified by a degree modifier such as hen very, as in (4) and (6), and whether they can be used in bicomparatives directly, as in (5) and (7). Nevertheless, Lin follows Doetjes (1997) in assuming that non-gradable verbs have a quantity argument, parallel to the degree argument of gradable adjectives. For instance, pao run has the semantics in (19a), where run (x)(d)(s) means x has done d-quantity of running in situation s, parallel to the semantics of adjectives like shengqi angry (19b). (18) Johni yue pao pro i yue kuai. run fast John ran faster and faster. (19) a. pao = x e d d s s run (x)(d)(s) <e, <d, <s, t>>> b. shengqi = x e d d s s angry (x)(d)(s) <e, <d, <s, t>>> If we incorporate this assumption into his analysis, the adverbial comparative in (18) (repeated from 10b) would have the structure in (20) and the interpretations in (21). (20) CP CP 1 CP 2 yue IP yue IP (21) a. pao = x e d d s s run (x)(d)(s) NP VP NP AP Zhangsan i pao pro i kuai run fast b. yue = P <d,<s, t>> g 1 g 2 s 1 s 2 [P(g 1 )( s 1 ) P(g 2 )( s 2 ) g 2 g 1 ] c. John i yue pao pro i yue kuai = g 1 g 2 s 1 s 2 [run (John)(g 1 )(s 1 ) run (John)(g 2 )(s 2 ) g 2 g 1 ] g 3 g 4 s 3 s 4 [s 1 s 3 s 2 s 4 R<<g 1, s 1 >, <g 3, s 3 >> R<<g 2, s 2 >, <g 4, s 4 >> fast (John)(g 3 )(s 3 ) fast (John)(g 4 )(s 4 ) g 4 g 3 ] (21c) reads as: For any pair of degrees g 1 and g 2, and any pair of situations s 1 and s 2 such that John has done g 1 -quantity of running in s 1, and John has done g 2 -quantity of running in s 2, and g 2 is greater than g 1, there exists a pair of degrees g 3 and g 4, and a pair of 45
situations s 3 and s 4 such that s 3 is an extended situation of s 1 and s 4 is an extended situation of s 2. John is fast to degree g 3 in s 3, and John is fast to degree g 4 in s 4. g 4 is greater than g 3. Moreover, g 1 in s 1 has a causative relation R relation with g 3 in s 3. g 2 in s 2 has a causative relation R relation with g 4 in s 4. In short, (21c) expresses a correlation between the quantity of John s running and his speed. However, given our discussion in the previous section, (21c) does not express the meaning of (15). Instead, it conveys the meaning of (22)(repeated from (10a). (22) John pao-de yue duo, ta (jiu) pao-de yue kuai. run-de much he (then) run-de fast The more John ran, the faster he went. In view of this flaw in his analysis, in the following section, we will provide a new analysis for yue yue which aims to capture the semantic difference between adverbial comparatives and comparative correlatives. 4. The Semantics of yue yue Let us start with preliminaries. We assume that gradable predicates (typically adjectives) contain a degree argument in their semantics, but lack a time argument; nongradable predicates (typically verbs) have a time argument, but lack a degree argument. Following this assumption, the non-gradable predicate like pao run has the interpretation in (23a), where run (x)(t)(s) reads as: x runs in situation s and at time t. It differs from the interpretation of gradable predicates like (23b) in that it does not contain a degree argument. (23) a. pao = x e t i s s run (x)(t)(s) <e, <i, <s, t>>> b. gaoxing = x e d d s s happy (x)(d)(s) <e, <d, <s, t>>> We propose that yue has two interpretations, as shown in (24a) and (24b). (24) a. yue = P <d, <s, t>> s 1 s 2 g 1 g 2 [P(g 1 )( s 1 ) P(g 2 )(s 2 ) g 2 g 1 ] b. yue = P <i, <s, t>> s 1 s 2 t 1 t 2 [P(t 1 )( s 1 ) P(t 2 )(s 2 ) t 2 t 1 ] (24a) is the interpretation of yue when it combines with a gradable predicate in comparative correlatives. This meaning essentially follows Lin s analysis of comparative correlatives in Mandarin Chinese. In (24a), yue takes a property of degrees P <d, <s, t>>, and a pair of situations s 1 and s 2. It returns a proposition which is true iff P is true of g 1 in s 1 and P is true of g 2 in s 2. g 2 is greater than g 1. The interpretation in (24b) is our proposed interpretation of yue when it is combined with a non-gradable predicate in adverbial comparatives. It minimally differs from (24a) in the type of the first argument that yue takes. P <i, <s, t>> in (24b) denotes a property of 46
times. Both degrees and times are orderable types, that is, allow an order to be defined on the elements in their type domain. The result of applying the meaning of yue in (24b) to the three arguments P <i, <s, t>>, s 1 and s 2, is a proposition true iff P is true of t 1 in s 1 and P is true of t 2 in s 2. t 2 temporally follows t 1. The dual interpretation of yue in (24a) and (24b) can successfully capture the semantic difference between the adverbial comparative in (10a) (repeated in 25) and the comparative correlative in (10b). Let us look at (25) first. (25) John yue pao yue kuai. run fast John ran faster and faster. Syntactically, we propose that (25) has a monoclausal structure, which is different from the biclausal structure of comparative correlatives. The evidence for this proposal comes from the following evidence. First some adverbial comparatives, which are structurally parallel to (25), do not allow an insertion of an overt subject and the morpheme jiu then in front of the second yue, while maintaining their original meanings. (26) Comparative Correlatives a. John yue shengqi, Mary jiu yue gaoxing. angry then happy The angrier John is, the happier Mary is. Adverbial Comparatives b. John i yue tiao, ta i jiu yue gao. jump he then fast (i)?? John jumps higher and higher. (ii) John becomes taller and taller from jumping. Second, comparative correlatives like (26a), which are clearly bi-clausal, allow an insertion of a future aspect marker in front of the second yue, and receive a future interpretation. However, if we do so with the adverbial comparative in (26b), the sentence receives a different meaning rather than just a future interpretation. (27) Comparative Correlatives a. John yue shengqi, Mary (jiu) hui yue gaoxing. angry then will happy The angrier John is, the happier Mary will be. 47
b. *John hui yue shengqi, Mary jiu yue gaoxing. will angry then happy Adverbial Comparatives c. John yue tiao, jiu hui yue gao. run then will fast. (i)?? John will jump higher and higher (ii) John will become taller and taller from jumping. d. John hui yue tiao yue gao. will run high John will jump higher and higher. Based on the above evidence, we propose that (25) has the LF in (28): (28) TP vp AdvP yue vp yue AdvP kuai NP VP fast Zhangsan i pao run The structure in (28) differs from Lin s structure in (20) in that (28) has a monoclausal structure. The predicate following the first yue pao run is the main predicate, and the predicate following the second yue kuai fast, is an adverb. The subject John is raised out of the vp to the spec of TP to receive a nominative case. Semantically, the vp in (28) denotes a set of temporally ordered situations in which John ran. The AdvP denotes a set of situations ordered based on John s running speed. The universal quantifier takes the vp and the AdvP as its semantic arguments and returns a proposition true iff John s speed increases over the temporally ordered running situations. The step-by-step interpretation of (28) is provided below: (29) a. pao = x e t i s s, run (x)(t)(s) b. John pao = t i s s run (John)(t)(s) 48
c. yue = P <i, <s, t>> s 1 s 2 t 1 t 2 [P(t 1 )( s 1 ) P(t 2 )(s 2 ) t 2 t 1 ] d. yue John pao = s1 s 2 t 1 t 2 [run (John)(t 1 )( s 1 ) run (John) (t 2 )(s 2 ) t 2 t 1 ] e. kuai = d d s s fast (d)(s) f. yue = P <d, <s, t>> s 1 s 2 g 1 g 2 [P(g 1 )( s 1 ) P(g 2 )(s 2 ) g 2 g 1 ] g. yue kuai = s 1 s 2 g 1 g 2 [fast (g 1 )( s 1 ) fast (g 2 )(s 2 ) g 2 g 1 ] h. = P <s, <s, t>> Q<s, <s, t>> s 1 s 2 [P(s 1 )( s 2 ) Q(s 1 )( s 2 )] i. yue Zhangsan pao yue kuai = s 1 s 2 [ t 1 t 2 [run (John)(t 1 )( s 1 ) run (John)(t 2 )(s 2 ) t 2 t 1 ] g 1 g 2 [fast (g 1 )( s 1 ) fast (g 2 )(s 2 ) g 2 g 1 ]] (29i) says that for any pair of situation s 1 and s 2, which are runnings by John, and such that s 2 is later than s 1, s 2 is faster than s 1 The comparative correlative in (10b), repeated below in (30), has a different interpretation. It means an increase of the quantity of John s running correlates with an increase of his speed. Let us calculate how this meaning is derived by incorporating the meaning of yue in (24a). (30) Johni pao-de yue duo, tai (jiu) pao-de yue kuai. run-de much he (then) run-de fast The more John ran, the faster he went. Syntactically, (30) has the biclausal structure in (31), following Lin (2007): (31) CP CP 1 CP 2 yue IP yue IP NP VP NP VP John i pao-de duo ta i pao-de kuai run-de fast he run-de fast 49
(32) a. pao g = x e s s run (x)(t)(s) b. John pao g = s s run (John)(t)(s) c. duo g = d d s s much (d)(s) d. John pao-de duo g = d d s s [run (John)(t)(s) much (d)(s)] e. yue g = P <d, <s, t>> s 1 s 2 g 1 g 2 [P(g 1 )( s 1 ) P(g 2 )(s 2 ) g 2 g 1 ] f. yue John pao-de duo = s 1 s 2 g 1 g 2 [[run (John)(t)( s 1 ) much (g 1 )( s 1 )] [run (John)(t)( s 2 ) much (g 2 )( s 2 ) ] g 2 g 1 ] g. ta i g = g(i) = John h. ta i pao g = s s run (John)(t)(s) i. kuai g = d d s s fast (d)(s) j. ta i pao-de kuai g = d d s s [run (John)(t)(s) fast (d)(s)] k. yue ta i pao-de kuai g = s 1 s 2 g 1 g 2 [[run (John)(t)( s 1 ) fast (g 1 )( s 1 )] [run (John)(t)( s 2 ) fast (g 2 )( s 2 ) ] g 2 g 1 ] l. g = P<s, <s, t> Q<s, <s, t> s1 s2 [P(s1)( s2) Q(s1)( s2)] m. yue John pao-de duo, ta yue pao-de kuai g = t s 1, s 2 g 1 g 2 [[run (John)(t)( s 1 ) much (g 1 )( s 1 )] [run (John)(t)( s 2 ) much (g 2 )( s 2 ) ] g 2 g 1 ] g 3 g 4 [[run (John)(t)( s 1 ) fast ( g 23 )(s 1 )] [run (John)(t)( s 2 ) fast ( g 4 )(s 2 )] g 4 g 3 ]] (32m) says: For any two situations s 1 and s 2 which are runnings by Zhangsan and such that the quantity of running in s 2 is greater than that in s 1, s 2 is also faster than s 1. So far we have seen how the proposed interpretations of yue account for the semantic difference between the adverbial comparative in (25) and the comparative correlative in (30). Before we conclude, some more explanations of adverbial comparatives are in order. First, though we have only examined the semantics of the monoclausal adverbial comparative in (25), adverbial comparatives can be biclausal as well. Below, let us take a brief look at some examples of biclausal adverbial comparatives. 50
(33) a. John yue pao, shengti yue jiankang. run body healthy As John was running, his body became healthier and healthier. b. John yue chang ge, xinqing yue hao. John sing songs mood good As John was singing, his mood became better and better. The examples in (33) are clearly biclausal. (33a) means that John s health improved over the time while he was running. It does not express a correlation between the quantity of John s running and his degree of healthiness, as shown by the scenario depicted in (34). (35) The scenario in which (33a) is intuitively true Time Mileage Degree of healthiness Day 3 Day 2 Day 1 John ran 2 miles John ran 1 mile John ran 3 miles 5 4 3 Neither does (33a) express a cumulative reading--a correlation between a running total of the quantity of John s running and his average degree of healthiness. As we have shown earlier (12&12 ), the cumulative reading of a comparative correlative usually has a weaker truth-condition than the temporal reading of an adverbial comparative. Second, the temporal reading of adverbial comparatives has a distinct status from the time reading that Lin (2007) has attributed to comparative correlatives like (35): (35) Tianqi yue re, wo jiu yue bushufu. weather hot I then uncomfortable The hotter the weather is, the more uncomfortable I feel. The meaning of (35) is represented by the formula in (36). It says: for all time pairs t 1 and t 2, if the weather is hotter at t 2 than it is at t 1,then I feel more uncomfortable at t 2 than at t 1. (36) t 1 t 2 [ d 1 d 2 [the weather is d 1 -hot at t 1 the weather is d 2 -hot at t 2 d 2 > d 1 ] d 3 d 4 [I am d 3 -comfortable at t 1 I am d 4 -comfortable at t 2 d 4 > d 3 ] Compare this meaning in (36) to the meaning of (37) in (38): 51
(37) John yue pao yue kuai. run fast John ran faster and faster. (38) s 1 s 2 [ t 1 t 1 2[John runs at t 1 in s 1 John runs at t 2 in s 2 ) t 2 t 1 ] d 1 d 2 [John s running is d 1 -fast in s 1 John s running is d 2 -fast in s 2 d 2 d 1 ]] Though both formulas make use of time variables, they have difference status. In (37), time variables are used as an ordering source such that situations are ordered temporally. In (36), degrees, rather than times, are used as an ordering source, such that times are ordered based on degrees rather than based on their temporal orderings. 5. Conclusion To conclude, in this paper, we have shown that yue yue in Mandarin can mark two types of comparatives the comparative correlative and the adverbial comparative, and these two types of yue yue comparatives are semantically distinct. The adverbial comparative has a necessary temporal reading, which is absent in the comparative correlative. We proposed that non-gradable predicates, mostly verbs, have a time argument, but no degree argument; gradable predicates, mostly adjectives, have a degree argument, but no time argument. We formulated two meanings for yue, depending on whether it combines with a gradable or non-gradable predicate. Our semantics for the comparative correlative maintains Lin s account, but it extends to the semantics of the adverbial comparative, which the previous analyses do not capture. Acknowledgement Various versions of this paper have been presented in Ruling5 at Rutgers, MACSIM1 at University of Pennsylvania, NACCL-22&IACL-18 at Harvard, TEAL-6 workshop in Beijing. We thank the audiences on all these occasions. We especially thank Roger Schwarzschild, Lisa Cheng, and Luther Liu for questions and comments. All remaining mistakes are our own. References Beck,S. 1997. On the semantics of comparative conditionals. Linguistics and Philosophy 20 (3), 229 271. [2] Erlewine, M. 2007. A New Syntax and Semantics for the Mandarin bi-comparatives. Master s thesis, University of Chicago. Doetjes, J. S. 1997. Quantifiers and Selection: On the Distribution of Quantifying Expressions in Fresh, Dutch and English. Doctoral dissertation, Leiden Unviersity. Hsiao, H.-Y. 2003. On proportional correlative constructions in Chinese and Mongolian. Journal of Taiwanese Language and Literature, 1:243-272. 52
Li, C and Thompson, S.1981. Mandarin Chinese: A functional reference grammar, University of California Press. Li. X. 2009. Degreeless Comparatives. Rutgers Dissertation Lin, J.W., 2007. On the semantics of comparative correlatives in Mandarin Chinese. Journal of Semantics 24 (2), 169 213. Lin, J (2009). Chinese comparatives and their implicational parameters. Natural Language Semantics 17: 1-27. Liu, C.S. L. 2008. The view from yue: Chinese comparative correlatives. Lingua 118, 1033-1061. Xiang, M. 2003. A phrasal analysis of Chinese comparatives. In Proceedings of the 39th Annual Meeting of the Chicago Linguistic Society. Xiang, M. 2005. Some Topics in Comparative Constructions. Doctoral dissertation, Michigan State University. 53